The ability of atmospheric data to resolve discrepancies in wetland methane estimates over North America.

Existing estimates of methane fluxes from North American wetlands vary widely in both magnitude and distribution. In light of these disagreements, this study uses atmospheric methane observations from the US and Canada to analyze seven different bottom-up, wetland methane estimates reported in a recent model comparison project. We first use synthetic data to explore how well atmospheric observations can constrain wetland fluxes. We find that observation sites can identify an atmospheric pattern from Canadian wetlands but not reliably from US wetlands. The network can also identify the spatial distribution of fluxes in Canada at multi-province spatial scales. Based upon these results, we then use real data to evaluate the magnitude, temporal distribution, and spatial distribution of each model estimate. Most models overestimate the magnitude of fluxes across Canada. Most predict a seasonality that is too narrow, potentially indicating an over-sensitivity to air or soil temperatures. In addition, the LPJ-Bern model has a spatial distribution that is most consistent with atmospheric observations. Unlike most models, LPJ-Bern utilizes land cover maps, not just remote sensing inundation data, to estimate wetland coverage. A flux model with a constant spatial distribution outperforms most other existing flux estimates across Canada. [ABSTRACT FROM AUTHOR]